image processing - Cropping out Extreme Distortion from a Homography -
i have picture of checkerboard taken arbitrary camera angle. find 2 vanishing points corresponding 2 sets of lines form checkerboard grid. these 2 vanishing points, compute homography checkerboard plane image plane.
i apply inverse homography re-render checkerboard top view. however, images, re-rendered top view large. is, due camera angle, inverse homography stretches parts of image (i.e. regions of image close 1 of vanishing points) large.
this takes unnecessarily large amount of memory, , of region becomes highly stretched stuff not need. so, when applying inverse homography, avoid rendering regions of image highly stretched. way this?
(i coding in matlab)
if need render checkerboard, without background, extract 4 corners of checkerboard , compute homography maps them 4 corners of square.
then can obtain rectified image of checkerboard warping input image homography, paying attention render needed region (ie square on map checkerboard).
Comments
Post a Comment